Health policy and systems research in access to medicines: a prioritized agenda for low- and middle-income countries

The WHO framework “Equitable Access to Essential Medicines: A Framework for Collective Action” [18] was used as a guide. This framework presents four dimensions of ATM: rational selection and use, affordable prices, sustainable financing, and reliable health and supply systems. Each of these four dimensions is influenced by determinants of ATM rooted in local, national and international context [19].

The stepwise approach used to determine research priorities was adapted from Ranson et al. [20, 21] and is summarised in Figure 1. First, in 17 countries across five regions, research teams performed an exhaustive literature review, focusing on ATM in their countries and regions, and which covered published and grey literature in local as well as international databases. This was followed by key informant interviews (KIIs) and definition of priority policy issues. This resulted in a series of country and regional reports, which were supplemented by KIIs with global stakeholders. Second, a literature scoping review was conducted to identify existing published research, which was then mapped against the priorities identified at country-, regional- and global-level. The findings of the first two steps were shared with a group of experts during a consultative workshop aimed at formulating priority research questions and ranking these in a prioritized HPSR agenda.

Seventeen countries were selected by the AHPSR based on the following criteria: i) the capacity of in-country researchers and institutions; ii) their ability to approach a diverse group of stakeholders in a meaningful dialogue; and iii) their ability to adopt a health system perspective on ATM. Efforts were made to keep a balance between regions and to include both low- and middle-income countries. The analytical framework [18] was presented and discussed in detail with research teams at an inception meeting. Meeting participants also discussed how best to design and implement data collection tools that would capture all elements of the framework and determinants of access at all levels of the health system. Country teams were given general directions for steps to follow in collecting and analysing data. They were asked to: i) perform a search of grey and published literature related to their country and region, and identify existing information on priority policy issues and existing research in ATM; ii) map relevant stakeholders from various groups: researchers, policy makers, health service providers, civil society and NGOs including patient representatives, donor agencies, pharmaceutical companies and other relevant stakeholders; iii) select a sample of stakeholders with adequate representation of all groups; perform KIIs to identify priority policy issues and research questions; iv) derive from these priority policy issues on ATM and provide a list of research questions if possible. Each team conceived search and survey methods adapted to their contexts and in local languages. Table 1 presents the list of countries by region, the number of informants in each country and whether identification of priority policy issues was based on consensus meeting vs. ranking (either in ranking meetings or as a component of KIIs).

Institutions were requested to seek ethical approval in their own country or region. Data collection took place between November 2010 and December 2011 and country/regional reports were submitted to the AHPSR between May and December 2011^b. The reports were read by two reviewers independently (MB and JH). Priority policy issues for each country were extracted and categorized according to the analytical framework. Results of independent data extraction were compared; discrepancies were discussed and resolved by returning to the original report. Cross-cutting priority policy issues appearing in at least three regions were identified and considered as being of high priority. Priority policy issues appearing in fewer than three regions were not considered as high priority.

During the data extraction process, some amendments were made to the original framework. The team observed that there was considerable overlap between issues related to domains “affordable prices” and “sustainable financing”. That is, when issues of affordability as a demand-side aspect were studied, naturally the supply-side, sustainable financing and implications, were also studied and vice versa. Thus, the two domains were merged into one, inclusive of both demand- and supply-side financing. Further, the domain “reliable health and supply system” covered a large number of policy issues and was therefore expanded and sub-domains defined. The final domains used were as follows:

To collect data on priority policy issues from a global perspective, 23 stakeholders from headquarters and regional offices, of international multilateral and bilateral agencies, academia and NGOs were interviewed. These informants were selected for their expertise and knowledge of ATM issues at international level. Priority policy issues identified at country- and regional-level^c were used to design a semi-structured telephone interview guide. In relation to each of the seven domains of the revised framework and through a series of open-ended questions, informants were asked whether they agreed with priority issues defined at country- and regional-level, and whether any issue was missing and would deserve to be prioritized. Interviews were performed between June and November 2011 by researchers external to the AHPSR with 23 informants who responded in their professional capacity. Interviews were conducted in English following email agreement and oral consent at the start of the interview. Interviews were not recorded but transcribed; transcripts were sent back to the interviewees for final agreement. Transcripts were read by two independent readers (JH, who performed the interviews and DJ, who did not) and priority policy issues identified by global key informants were extracted in an Excel spread sheet, according to the modified analytical framework. Results of independent reviews were compared and discrepancies were discussed and resolved by returning to the original transcripts.

Policy issues were categorized according to the seven domains above. Topics were ordered firstly according to the number of regions in which the issue was raised and second according to the number of interviewees who raised the issue at global level.

Using the list of priorities identified through the KIIs and the country reports, a search strategy was devised to extract relevant literature. The search strategy was adapted from a review recently published by the AHPSR [11] and is presented in Additional file 1 and described below. As PubMed is limited in its inclusion of relevant LMIC literature, additional sources were considered. Country and regional teams conducted literature reviews on ATM in their local context, using international and local databases, resulting in a comprehensive reference list for each country and region. To supplement the global search, reference lists provided by country and regional teams were included to strengthen the inclusiveness of the literature scoping (see Additional file 2 for the list of databases searched by country, regional and global research teams). Finally, a report by Gray and Suleman as commissioned by the AHPSR was used as well to further broaden the search [22]. These additional sources covered major databases for published and grey literature and databases specific to LMICs. From these, relevant publications and systematic reviews were included in the scoping exercise.

A global stakeholders meeting was organized in Bangkok on 6–8 March 2012. Participants from academia were selected based on their records of research and publications in the field of pharmaceuticals and health systems. Non-academia participants were selected to represent decision makers in the public and private sectors, which included multi-lateral agencies, government agencies, NGOs, pharmaceutical companies, and private foundations. The 20 participants included three who lead country-level exercises and five who were interviewed as part of the global stakeholders’ interviews. Participation was voluntary and based on availability. In preparation for the meeting, participants received summaries of individual country- and regional-level priority setting exercises, a background paper detailing the methods and results of country-, regional- and global-level priority setting as well as methods and results of literature scoping, and a spreadsheet presenting the full list of references including abstracts. During the first half of the meeting, 14 participants from academia reviewed the list of priority policy issues resulting from country-, regional-, and global-level research; and agreed on the categories as well as the wording. Once agreement was reached, they formulated research questions corresponding to each of the ATM policy issues identified in the first two steps. This exclusive time for formulation of research questions was designed to overcome the difficulty of translating policy issues into health policy and systems research questions, which required the expertise of researchers. During the second half of the meeting, non-academic participants joined the group, reviewed the list of policy issues, and drafted questions formulated by academia. Consensus was reached on a final set of 18 unranked research questions.

Participants also debated and reached consensus on four ranking criteria, to which they decided to attribute the same weight:

In addition, participants suggested and agreed to add an overall score, irrespective of ranking criteria, that would assess the overall importance of each research question in LMICs and proposed that each participant lists his/her top five priority research questions at the end of the questionnaire. The questionnaire was modified to account for these suggestions.

Participants ranked the 18 research questions through an individual anonymous questionnaire. A five-point Likert scale was used for each ranking criteria where one represented low priority and five represented high priority. Three different methods were used for combining scores across participants:

First, the “total score” involved: i) for each participant and question, summing the four ranking criteria (given the consensus decision to assign equal weight to each criterion); and ii) taking the flat (non-weighted) average of scores for each question across all participants. Questions were ranked from one (top rank, highest average total score) to 18 (lower rank, lowest average total score).

Second, the “overall score” was entered also using a five-point Likert scale where one represented a low overall priority and five represented a high overall priority. The overall score was also combined by taking a flat average for each question, across all participants. Questions were ranked from one (top rank, highest average overall score) to 18 (lower rank, lowest average overall score).

Third, the “top five” was a simple count of the number of times each research question was listed in the top five priorities of all participants and questions were ranked from one (highest number of counts) to 18 (lowest number of counts).

The results of the ranking exercise were analysed on site, presented and discussed in plenary, and consensus was reached on top priority questions.

A total of 26 crosscutting high priority policy issues were identified at country- and regional-level. Nine policy issues were consistently raised in all five regions and related to: i) perception of generic medicines by patients, prescribers and providers, and their preference for branded medicines; ii) knowledge, awareness, and health seeking behaviour of patients; iii) development of central and essential policy documents such as standard treatment guidelines and national essential medicine lists; iv) universal coverage, social health protection and coverage of medicines under health financing arrangements; v) level of public and government funding for health and medicines; vi) inefficiencies in budget disbursement and implementation; vii) medicine pricing policies; viii) capacity of staff and systems for procurement, supply and stock management; and ix) fight against counterfeit medicines. An additional eleven issues were raised in four regions and six issues were raised in three regions. Policy issues raised by less than three regions are considered as low priority and are not presented here.

A total of 23 informants were interviewed at global level, of which six were women. Ten informants originated from a LMIC. Interviewees had expertise either in pharmaceuticals or health systems and worked at headquarters of international multilateral agencies (n = 8), at regional offices of multilateral agencies (n = 4), in a bilateral donor organization (n = 1), in international NGOs (n = 4) and six were researchers in academic institutions. The results of global KIIs were fairly consistent with country- and regional-level results. Global KIIs placed more emphasis on the role of human resources for health in medicine access, although for both groups, the specific policy focus was the impact of financial and non-financial providers’ incentives on access. The second most important domain for global key informants was the same as for countries and regions: sustainable financing and affordable prices. Consistent with data from countries and regions, quality issues came third with global key informants; however, here, a focus on substandard medicines was considered to be more critical than counterfeit medicines as these were seen as a symptom of other system failures outside of quality assurance. In general, all priority policy issues that were identified consistently in all five regions were presented as important concerns for global key informants.

A consolidation of results from country, regional, and global exercises allowed authors to define a final list of 18 crosscutting high-priority policy issues related to ATM in LMICs. Inclusion of priority policy issues in this list was based first on the number of regions where the issue was raised and second according to the number of informants who raised the issue (Table 2). As the results of Global KII were consistent with country- and regional-level results, no additional policy issues raised by Global Key Informants were added to the list of country and regional priorities. A few policy issues among the list of 26 priorities resulting from country- and regional-level exercises were combined as they appeared redundant or strongly linked (e.g., “provider payment methods and financial incentives to providers” was combined with “financial and non-financial incentives to providers”; “procurement and supply of generic medicines” was combined with “quality of generic medicines”).

A total of 1,279 articles were retrieved from PubMed; 619 were found relevant for our study. Results show a distribution of publications predominantly in Sub-Saharan Africa (32.1%), followed closely by the Asian continent (23.6%, with 12.6% in South Asia and 11% in East Asia and the Pacific), and Latin America and the Caribbean (18.6%). A majority of publications focused on the public sector (65.4%) followed by both public and private sector (8.7%); 15.8% of publications studied only the private sector. Half of the publications had a specific urban focus and 12% specifically targeted rural settings.

From reference lists of country and regional priority-setting reports, an additional 72 articles were identified as relevant to the search. Sixteen systematic reviews were also extracted from the review by Gray and Suleman [22]. A total of 707 references were categorized, based on screening through each article, according to the list of priority policy issues in ATM (Table 3).

Priority policy issues for which there are fewer references in the literature include resources allocation for health and medicines; deployment and shortage of human resources for health; counterfeit medicines; procurement, supply and stock management; and crosscutting policies outside the health sector affecting health and medicines. Moreover, some of the priorities are only covered by less robust research methodologies. For example, for the following five priorities, no publication was identified that attempts to assess the effects or outcomes of specific interventions: resource allocation for health and medicines; health seeking behaviour of patients, households and communities; crosscutting policies outside the health sector affecting health and medicines; governance over the private sector and financial and non-financial incentives for providers.

Fourteen academic researchers were present during the first half of the meeting, including one private foundation and three researchers who had performed country- and regional-level priority setting exercises. They came from institutions in Brazil, China, Ghana, Iran, Mexico, South Africa, Switzerland, Sweden, Thailand, and the United States of America. Seven stakeholders joined the academic group during the second half of the meeting; they came from international multilateral agencies (n = 2), a government agency (n = 1), NGOs (n = 2), a pharmaceutical company (n = 1), and a private foundation (n = 1). Two WHO officers were also present. Eight participants were female. Research questions addressing the 18 priority policy issues were formulated by the academic researchers, and then reviewed by the second group of stakeholders who introduced minor editorial revisions. While the experts did have the option of developing more than one research question per issue, they jointly decided on a single question corresponding to each policy issue. Table 4 presents the research questions formulated by the group and the results of the ranking exercise according to the three ranking methods. It also disaggregates the results according to the four ranking criteria (Innovation, Impact, Equity, and Lack of Research).

Participants in the ranking exercise were presented with preliminary quantitative analyses of all three ranking methods. They reached a consensus that priority should be given to research questions 1, 12 and 18.

Question 1 (“In risk protection schemes, which innovations and policies improve equitable access to and appropriate use of medicines, sustainability of the insurance system, and financial impact on insurance members?”) was overwhelmingly ranked as a top priority, irrespective of the ranking method used. This question also scored as a high priority according to all ranking criteria (top priority with regards to Impact and Lack of Research, second priority with regards to Innovation and Equity). This research topic corresponds to a priority policy issue raised consistently in all five regions and by 18 out of 23 global key informants (Table 2). As confirmed by our literature scoping, medicine publications have only recently shifted towards examining health-financing arrangements and investigating medicine prices in LMICs. Our priority setting exercise confirms an on-going interest in these topics, which are also consistent with the current trend of focussing on universal coverage in a post-Millennium Development Goal world.

The top five ranking method placed Question 18 (“How can stakeholders use the information available in the system, e.g., price, availability, quality, utilization, registration, and procurement, in a transparent way towards improving access and use of medicines?”) in second position while the results of the total score and overall score methods pointed to Question 12 (“How do policies and other interventions into private markets (such as information, subsidies, price controls, donation, regulatory mechanisms, promotion practices, etc.) impact on access to and appropriate use of medicines?) as a second priority. Question 12 was considered highly innovative and addressing a substantial research gap, as confirmed by our literature scoping. Question 18, related to medicine information, was regarded as potentially achieving high impact.

Consensus was reached among the meeting participants for the above top three ranking research questions that needed attention and investment in the future. Participants suggested that the results of the ranking, disaggregated by ranking criteria, be made publicly available to allow researchers, decision makers and funders to select research questions according to their expertise and their needs, depending on whether they wanted to focus on equity, innovation, impact on health and health systems or on filling existing research gaps.